There is no solution without a solution to Feiganbaum's ( and Minsky) third problem. Not anyone, Hinton, LeCun, Pinchar or Altman propose a solution.
Boggles the mind to think this problem is solved by statistics. In fact the the entire AI debacle is due to not knowing how to solve the problem. No free lunch.
First is the problem of knowledge representation. How shall the knowledge of the
field be represented as data structures in the memory of the computer, so that they
can be"conveniently accessed for problem-solving?
Second is the problem of knowledge utilization. How can this knowledge be used
Essentially,
in problem solving? Essentially, this is the question of the design of the inference engine. What designs for
What designs for the inference engine are available?
Third, and most important, is the question of knowledge acquisition.
How is it
possible
to acquire the knowledge so important for problem-solving automatically or at least semi-automatically, in a way in which the computer facilitates the transfer of expertise from humans (from practitioners or from their texts or their data) to the symbolic data structures that constitute the knowledge representation in the
acquisition
machine?
Knowledge acquisition is a long-standing problem of Artificial
Intelligence. For a long time it was cloaked under the word "learning". Now we are able to be more precise about the problem of machine learning; and with this increased precision has come a new term,
"knowledge acquisition research".
This is the most important of the central problems of Artificial Intelligence research. The reason is simple: to enhance the performance of AI's programs,
the inference
knowledge is power. The power does not reside in the inference procedure.
It was solved 25 years ago. National security was the reason for stealth. Epistemic disaster is the reason to build immutable knowledge.
“Boggles the mind to think this problem is solved by statistics“
As the great mathematician John von Neumann once said (according to Enrico Fermi as related by Freeman Dyson)
“With 4 parameters, I can fit an elephant, with 5, I can make him wiggle his trunk and with billions, I can make him hallucinate pink elephants like a Large Language Model”
A crow with a brain the size of a nut, can still show more intelligence than these LLM monstrosities. Instead of using future e-waste and a billion jigawattz of electricity, it works for the equivalent of peanuts and the occasional shiny scrap of foil.
Did you see the robot butler that came out recently? It tried to gently pet a walnut open and moved like an elderly person on booze. Crow would have got the job done.
I'm not sure why people keep looking for intelligence in a glorified thesaurus. I've never seen an encyclopedia jump off the shelf and crack a walnut or solve physics. Encyclopedia are hand edited to be maximally truthful, perhaps all that's missing is a charging cable. 🤔
Lol yes! 😀 I'm still trying to come to terms with quants, VCs, and CEOs believing that the LLM will replace a bunch of workers, while at the same time being extra adverse to hiring mentally handicapped people.
At a high level, if you don't look too deep, an LLM is like a brain with all the parts cut off except language processing, and then that bit is hit with a mallet and microwaved for a bit.
I mean, I guess I don't expect the big swinging dicks in the room to be using their big brains... maybe the Epstein files have a much broader reach than I anticipated and they'll have to redact photos of the island's server room for being too sticky or whatever. 🤷♂️ Have you seen how hard they're getting over data centers with no hope of profitability? They're like dudes on OnlyFans listening to tech bros promise to whip out the good stuff once the tip jar is full. 😀
Hello Gary: You don't need to comment on a prior note (see my comment in your last blogpost) but I wanted to be sure you at least knew about the article and link I posted there from NATURE online magazine with interviews with six people in AI. Thank you for keeping us up on things in the field, however. The foul ethics situation will not go away anytime soon.
Gemini 3 is good for one thing - actually being able to search well without the useless hash Google has become. I know two things pretty well in this world; trading and analog photography. Don't ask. I do use both to see what LLMs do as I can catch them out more easily. Gemini took a problem and recommended a reckless no-undo course of action. When I prompted that this was unnecessary and a one-way street, it noted the lapse. What happens to users who are not sceptical?
The LLM race is reminiscent of the “race” between horses on a merry-go-round.
“The AI-go-round”
Please stop the ride!
And set me aside!
I’ve had it with chats
And had it with bots
And had it with stats
And benchmarks and plots
I’m dizzy as hell
As any will tell
I long for the ground
Not AI-go-round
“The AI-go-round (2)”
Google’s in the lead
By roughly half the race
But Open AI’s steed
Is keeping with the pace
And if we look again
When Google’s horse is last
Then Open AI’s win
Will surely come to pass
There is no solution without a solution to Feiganbaum's ( and Minsky) third problem. Not anyone, Hinton, LeCun, Pinchar or Altman propose a solution.
Boggles the mind to think this problem is solved by statistics. In fact the the entire AI debacle is due to not knowing how to solve the problem. No free lunch.
Al_ Knowledge_Origins_175986... 1980
https://stacks.stanford.edu/file/druid:cn981xh0967/cn981xh0967.pdf
First is the problem of knowledge representation. How shall the knowledge of the
field be represented as data structures in the memory of the computer, so that they
can be"conveniently accessed for problem-solving?
Second is the problem of knowledge utilization. How can this knowledge be used
Essentially,
in problem solving? Essentially, this is the question of the design of the inference engine. What designs for
What designs for the inference engine are available?
Third, and most important, is the question of knowledge acquisition.
How is it
possible
to acquire the knowledge so important for problem-solving automatically or at least semi-automatically, in a way in which the computer facilitates the transfer of expertise from humans (from practitioners or from their texts or their data) to the symbolic data structures that constitute the knowledge representation in the
acquisition
machine?
Knowledge acquisition is a long-standing problem of Artificial
Intelligence. For a long time it was cloaked under the word "learning". Now we are able to be more precise about the problem of machine learning; and with this increased precision has come a new term,
"knowledge acquisition research".
This is the most important of the central problems of Artificial Intelligence research. The reason is simple: to enhance the performance of AI's programs,
the inference
knowledge is power. The power does not reside in the inference procedure.
It was solved 25 years ago. National security was the reason for stealth. Epistemic disaster is the reason to build immutable knowledge.
http://intellisophic.net/2025/09/12/the-fundamental-innovation-orthogonal-corpus-indexing-oci/
“Boggles the mind to think this problem is solved by statistics“
As the great mathematician John von Neumann once said (according to Enrico Fermi as related by Freeman Dyson)
“With 4 parameters, I can fit an elephant, with 5, I can make him wiggle his trunk and with billions, I can make him hallucinate pink elephants like a Large Language Model”
A crow with a brain the size of a nut, can still show more intelligence than these LLM monstrosities. Instead of using future e-waste and a billion jigawattz of electricity, it works for the equivalent of peanuts and the occasional shiny scrap of foil.
Did you see the robot butler that came out recently? It tried to gently pet a walnut open and moved like an elderly person on booze. Crow would have got the job done.
I'm not sure why people keep looking for intelligence in a glorified thesaurus. I've never seen an encyclopedia jump off the shelf and crack a walnut or solve physics. Encyclopedia are hand edited to be maximally truthful, perhaps all that's missing is a charging cable. 🤔
Crows are highly intelligent.
If you asked a crow what it thought of LLMs, it would undoubtedly say “caca(w)”.
Lol yes! 😀 I'm still trying to come to terms with quants, VCs, and CEOs believing that the LLM will replace a bunch of workers, while at the same time being extra adverse to hiring mentally handicapped people.
At a high level, if you don't look too deep, an LLM is like a brain with all the parts cut off except language processing, and then that bit is hit with a mallet and microwaved for a bit.
I mean, I guess I don't expect the big swinging dicks in the room to be using their big brains... maybe the Epstein files have a much broader reach than I anticipated and they'll have to redact photos of the island's server room for being too sticky or whatever. 🤷♂️ Have you seen how hard they're getting over data centers with no hope of profitability? They're like dudes on OnlyFans listening to tech bros promise to whip out the good stuff once the tip jar is full. 😀
“Google’s commitment to operate sustainably”?
There is nothing even remotely sustainable about any of the stuff going on in the current botosphere.
Botsponential growth is like cancer.
If Google were to make those TPUs commercially available at scale, would they become “CheaPUs”?
Hello Gary: You don't need to comment on a prior note (see my comment in your last blogpost) but I wanted to be sure you at least knew about the article and link I posted there from NATURE online magazine with interviews with six people in AI. Thank you for keeping us up on things in the field, however. The foul ethics situation will not go away anytime soon.
Gemini 3 is good for one thing - actually being able to search well without the useless hash Google has become. I know two things pretty well in this world; trading and analog photography. Don't ask. I do use both to see what LLMs do as I can catch them out more easily. Gemini took a problem and recommended a reckless no-undo course of action. When I prompted that this was unnecessary and a one-way street, it noted the lapse. What happens to users who are not sceptical?
Google CEO Sundar Pichai had just better hope Gemeni 3 didn’t have a job interview at 3:30.
fwiw the tpu's are available on google's cloud.